Study On Registration Algorithm Of Swarm Intelligence Optimization For Point Cloud Based On Curvature Information

With the continuous development of information technology and the continuous improvement of image scanning equipment,three-dimensional image digitization technology has been gradually applied in many fields such as industrial development,medical treatment,product quality detection and virtual scene construction with its advantages of high efficiency and low cost.For the three-dimensional point cloud registration,the existing registration technology still has some problems,such as low registration accuracy,low efficiency and it has high requirement for environmental conditions.There is still some room for further exploration.In view of the above problems,the main work carried out in this paper is as follows:Firstly,to accelerate the speed of convergence caused by large amount of points,An new point cloud registration method based on curvature information and the artificial bee colony algorithm is proposed in this paper.For point cloud registration,it has the problems of large amount of computation and slow convergence speed to only use swarm intelligence optimization algorithm.In this paper,the curvature information of point cloud is introduced during the feature extraction,and part of the curvature feature points are used to participate in the registration process.In the process of population optimization,the range of finding the corresponding points is restrained by curvature information.And the size of point clouds involved in the calculation is smaller.So the amount of calculation in the optimization process is reduced.Some experiments show that,compared with the registration algorithm only using random point selection method and point cloud spatial coordinate information,the proposed algorithm can effectively accelerate the convergence speed of registration without reducing the registration accuracy,and significantly shorten the time of point cloud registration.Secondly,the selection strategy of the registration point cloud sampling point extraction stage is discussed and analyzed.The number of curvature feature points sampled in point cloud is small and the information of point cloud is insufficient,which may affect the efficiency and accuracy of registration.An improved point cloud registration algorithm based on curvature information and artificial bee colony algorithm is proposed.That means that every time the population passes through a certain optimization algebra,it updates the set of curvature sampling points to re-select points.On the premise that the number of sampling points is not increased,the point cloud information of curvature used in registration process is increased.Experiments show that the proposed algorithm improves the registration accuracy and efficiency.